Vascular Endothelial Growth Factor-B Acts as a Coronary Growth Factor in Transgenic Rats Without Inducing Angiogenesis, Vascular Leak, or Inflammation

Maija Bry, BMed; Riikka Kivela¨, PhD*; Tanja Holopainen, MD*; Andrey Anisimov, PhD; Tuomas Tammela, MD, PhD; Jarkko Soronen, MSc; Johanna Silvola, MSc; Antti Saraste, MD, PhD; Michael Jeltsch, PhD; Petra Korpisalo, MD; Peter Carmeliet, MD, PhD; Karl B. Lemstro¨m, MD, PhD; Masabumi Shibuya, MD, PhD; Seppo Yla¨-Herttuala, MD, PhD; Leena Alhonen, PhD; Eero Mervaala, MD, PhD; Leif C. Andersson, MD, PhD; Juhani Knuuti, MD, PhD; Kari Alitalo, MD, PhD

Background—Vascular endothelial growth factor-B (VEGF-B) binds to VEGF receptor-1 and neuropilin-1 and is abundantly expressed in the heart, skeletal muscle, and brown fat. The biological function of VEGF-B is incompletely understood. Methods and Results—Unlike placenta growth factor, which binds to the same receptors, adeno-associated viral delivery of VEGF-B to mouse skeletal or heart muscle induced very little angiogenesis, vascular permeability, or inflammation.

As previously reported for the VEGF-B167 isoform, transgenic mice and rats expressing both isoforms of VEGF-B in the myocardium developed cardiac hypertrophy yet maintained systolic function. Deletion of the VEGF receptor-1 domain or the arterial endothelial Bmx tyrosine kinase inhibited hypertrophy, whereas loss of VEGF-B interaction with neuropilin-1 had no effect. Surprisingly, in rats, the heart-specific VEGF-B transgene induced impressive growth of the epicardial coronary vessels and their branches, with large arteries also seen deep inside the subendocardial myocardium. However, VEGF-B, unlike other VEGF family members, did not induce significant capillary angiogenesis, increased permeability, or inflammatory cell recruitment. Conclusions—VEGF-B appears to be a coronary growth factor in rats but not in mice. The signals for the VEGF-B–induced cardiac hypertrophy are mediated at least in part via the endothelium. Because cardiomyocyte damage in myocardial ischemia begins in the subendocardial myocardium, the VEGF-B–induced increased arterial supply to this area could have therapeutic potential in ischemic heart disease. (Circulation. 2010;122:1725-1733.) Key Words: angiogenesis Ⅲ coronary disease Ⅲ hypertrophy

oronary artery disease leads to compromised myocardial leakage, inflammation, and the formation of angioma-like Cblood supply and the typical symptoms of stress-induced vascular structures,4–7 which has hampered its utility in angina. Although pharmaceutical therapy and revasculariza- therapeutic angiogenesis. Among the VEGF family members, tion of stenotic epicardial coronary arteries are the standard placenta growth factor (PlGF) has shown the most promise therapy for coronary artery disease, many patients with for promoting therapeutic arteriogenesis in preclinical advanced disease or small-vessel disease respond poorly to studies.7 these treatments. Novel therapeutic strategies for promoting collateral artery formation, or arteriogenesis, are in high Clinical Perspective on p 1733 demand.1,2 Although vascular endothelial growth factor VEGF-B, isolated in 1995,8 has been an exceptional (VEGF) is the most potent angiogenic factor for possible member of the VEGF family in that efforts to discover a therapy of myocardial ischemia,3 it also promotes vascular blood vascular function for VEGF-B have had largely nega-

Received March 29, 2010; accepted August 18, 2010. From the Molecular/Cancer Biology Laboratory and Institute for Molecular Medicine Finland (M.B., R.K., T.H., A.A., T.T., J. Soronen, M.J., K.A.) and Department of Pathology (L.C.A.), Haartman Institute, Biomedicum Helsinki, University of Helsinki and Helsinki University Central Hospital, Helsinki, Finland; Turku PET Centre, Turku University Hospital, Turku, Finland (J. Silvola, A.S., J.K.); Institute of Biomedicine, Pharmacology, University of Helsinki, Helsinki, Finland (E.M.); Biotechnology and Molecular Medicine, A.I. Virtanen Institute for Molecular Sciences, Biocenter Kuopio, University of Eastern Finland, Kuopio, Finland (P.K., S.Y.-H., L.A.); Department of Molecular Oncology, Tokyo Medical and Dental University, Tokyo, Japan (M.S.); Cardiopulmonary Research Group, Transplantation Laboratory, University of Helsinki and Department of Cardiothoracic Surgery, Helsinki University Central Hospital, Helsinki, Finland (K.B.L.); and Vesalius Research Center, KU Leuven, Leuven, Belgium (P.C.). *Drs Kivela¨ and Holopainen contributed equally to this article. The online-only Data Supplement is available with this article at http://circ.ahajournals.org/cgi/content/full/CIRCULATIONAHA.110.957332/DC1. Correspondence to Kari Alitalo, Molecular/Cancer Biology Laboratory, Biomedicum Helsinki, PO Box 63 (Haartmaninkatu 8), FI-00014 Helsinki, Finland. E-mail [email protected] © 2010 American Heart Association, Inc. Circulation is available at http://circ.ahajournals.org DOI: 10.1161/CIRCULATIONAHA.110.957332 1725 1726 Circulation October 26, 2010 tive results.6 VEGF-B knockout (KO) mice are viable and Blood Pressure Measurements display at most mild cardiac phenotypes such as a slightly and Echocardiography smaller heart size in 1 genetic background or a mild delay in Blood pressure was measured with the CODA Non-Invasive Blood atrioventricular coupling in another strain.9,10 Transgenic Pressure System for Mice and Rats (Kent Scientific Corp, Tor- (TG) or adenoviral overexpression of VEGF-B in the skin or rington, Conn). Transthoracic echocardiography was performed with an Acuson Sequoia 512 Ultrasound System and an Acuson Linear skeletal muscle increased blood vessel density only minimal- 15L8 transducer (Siemens Medical Solutions, Mountain View, ly.11,12 Interestingly, however, strong overexpression of Calif). VEGF-B by adenoviral transduction induced heart-specific capillary dilation and increased collateral blood vessel growth Micro–Computed Tomography Imaging of the after myocardial infarction in pigs.13 Cardiac Vessels VEGF-B is expressed as 2 RNA splice isoforms.14 Recently, Coronary angiographies were performed with the Inveon micro– we demonstrated that when overexpressed in the mouse heart, computed tomography scanner (Siemens, Knoxville, Tenn). The ascending aorta was cannulated and clamped, and iodinated intra- the heparin-binding VEGF-B167 isoform induced cardiac hyper- trophy without overt angiogenesis, although it caused mild vascular contrast agent eXIATM160XL (Binitio Biomedical Inc, Ottawa, Ontario, Canada) was carefully injected manually to fill the 12 enlargement of myocardial blood capillaries. The other iso- cardiac blood vessels, avoiding very high pressure. form, VEGF-B186, is O-glycosylated, proteolytically processed, and freely diffusible. Both isoforms bind to and activate VEGF Assessment of Myocardial Perfusion, Oxygen receptor-1 (VEGFR-1) and neuropilin-1.15,16 Several studies Consumption, and Efficiency of Work have indicated that the VEGF-B isoforms are expressed simul- Eight rats were given a slow bolus of 30Ϯ24 MBq of [11C]acetate 14 taneously in various tissues. PlGF binds to the same receptors and imaged with the Inveon micro–positron emission tomography as VEGF-B and has been shown to promote angiogenesis and scanner (Siemens, Knoxville, Tenn). arteriogenesis in pathological conditions.17 The aim of this study was to investigate the mechanisms Statistical Analysis of VEGF-B action in the heart in more depth. A critical Values are presented as meanϮSD unless otherwise indicated. question about the effects of VEGF-B in the myocardium Statistical analysis was performed with 1-way ANOVA (posthoc is how they differ from those of PlGF and whether with Tukey test) or with the 2-tailed unpaired Student t test unless VEGF-B could provide a means to improve myocardial otherwise specified in the Results. Differences were considered Ͻ function in the failing heart. Because adenoviral transduc- statistically significant at P 0.05. tion in an immunocompetent host is short-lived and results in nonspecific inflammation, we have instead used adeno- Results associated viral and TG delivery of VEGF-B to the mouse Unlike PlGF, VEGF-B Fails to Induce Capillary and rat heart. Angiogenesis or Arteriogenesis in Mouse Skeletal or Cardiac Muscle Methods To compare the effects of VEGF-B and PlGF, which bind to A detailed description of the methods used is provided in the VEGFR-1 and neuropilin-1, we injected recombinant AAVs online-only Data Supplement. encoding VEGF-B167, VEGF-B186, and PlGF, as well as VEGF and human serum albumin as positive and negative Construction and Preparation of the Recombinant controls, respectively, into mouse tibialis anterior muscles. Immunofluorescence staining of the muscles for platelet Adeno-Associated Virus Vectors endothelial cell adhesion molecule (PECAM)-1 and Mouse VEGF , PlGF, VEGF-B , VEGF-B , and human serum 120 167 186 ␣ albumin complementary DNAs were cloned into the psubCMV- -smooth muscle actin (SMA) 4 weeks after injection indi- WPRE recombinant adeno-associated virus (AAV) expression vec- cated strong capillary angiogenesis with primarily smooth tor. AAV particles were injected into tibialis anterior muscles or the muscle cell–coated vessels in PlGF-injected muscles, left ventricle. whereas no evidence of angiogenesis was seen in muscles overexpressing either VEGF-B isoform (Figure 1). In the Generation of ␣MHC–VEGF-B TG Mice and Rats VEGF-injected muscles, endothelial cell proliferation resem- A fragment of the human VEGF-B and the mouse VEGF- bled angioma formation (Figure 1). As expected on the basis ␣ BEx1–5 fragment were isolated and cloned into the -myosin heavy of prior work with adenoviral vectors, PlGF and VEGF ␣ chain ( MHC) promoter expression vector (a kind gift from Dr increased blood perfusion in the muscles, whereas even a Jeffrey Robbins). TG animals were generated by microinjection of fertilized oocytes from FVB/N mice or HsdBrl:WH Wistar rats. All combination of VEGF-B167 and VEGF-B186 had no effect on animal experiments were approved by the Provincial State Office of perfusion (Figure IA in the online-only Data Supplement). Southern Finland and carried out in accordance with institutional Interestingly, Evans Blue dye injection experiments indicated guidelines. that although PlGF and VEGF increased vascular permeabil-

ity, VEGF-B167 and VEGF-B186 did not (Figure IB in the Immunohistochemistry, Microscopy, and online-only Data Supplement). Similar effects were obtained Image Analysis when the same AAV vectors were expressed in the myocar- The antibodies and methods used are detailed in the online-only Data dium (Figure II in the online-only Data Supplement). Impor- Supplement. tantly, mice injected with PlGF or VEGF vectors had to be Bry et al VEGF-B as a Coronary Growth Factor 1727

approach. Because several studies have indicated that both VEGF-B isoforms are simultaneously expressed,14 we over- expressed the full-length human VEGF-B gene under the ␣MHC promoter in mice and rats. The schematic structure of the transgene is shown in Figure IVA in the online-only Data Supplement. Analysis of the TG mice and rats at 3 to 6 months of age indicated robust transgene expression in Western blot analysis of the heart lysates (Figure IVC in the online-only Data Supplement and data not shown). As can be seen from that figure, all 5 TG rat founder hearts contained

both full-length and processed VEGF-B186 polypeptides (32 and 14 kDa, respectively), as well as VEGF-B167 polypep- tides (22 kDa). Immunofluorescence staining for VEGF-B in the rat hearts demonstrated a mosaic pattern of expression in the cardiomyocytes (TG2), except for the highest expressing founder (TG3), which showed staining in all cardiomyocytes (Figure IVE in the online-only Data Supplement). Three of these founder lines (TG2 through TG4) were used in subse- quent experiments with essentially similar results. Both the TG mice and rats developed cardiac hypertrophy, with an increased ratio of heart to body weight and cardio- myocyte size as determined by staining of the cardiomyocyte plasma membrane with anti-dystrophin antibodies at 4 to 5 months of age (Figure 2A through 2C). Echocardiographic analysis of the TG rat hearts revealed a significantly in- creased left ventricular wall thickness at 5 months of age (Figure 2D and the Table), which was associated with a decrease in end-systolic volume (the Table). Interestingly, however, the TG rats maintained heart function, as shown by Figure 1. Comparison of the effects of VEGF-B and PlGF analysis of the ejection fraction and fractional shortening (the expressed in skeletal muscle via AAV vector delivery. A, Repre- Table). In addition, the TG rats tended to have lower blood sentative PECAM-1– (green) and SMA- (orange) stained sections of tibialis anterior muscles injected with AAVs encoding VEGF- pressure and heart rate than wild-type (WT) controls (Figure

B167, VEGF-B186, PlGF, VEGF, or human serum albumin (HSA). 2E and the Table). No degenerative changes comparable to B, Quantification of the total PECAM-1– and SMA-positive sur- those seen previously in ␣MHC–VEGF-B mouse hearts12 ϭ ␮ Ͻ 167 face areas in muscle sections. Scale bar 100 m. *P 0.05. were seen in the rat cardiomyocytes at 4 months of age in electron microscopy or in older rats, whereas cardiomyocyte euthanized within 2 weeks because of edema, whereas the damage and fibrosis were evident in light microscopy of VEGF-B–expressing mice showed no signs of distress (data 1-year-old ␣MHC–VEGF-B mice (Figure V in the online- not shown). only Data Supplement and data not shown).

VEGF-B, in Contrast to PlGF, Does Not Induce The VEGF-B Transgene Induces Strong Coronary Inflammation in Skeletal or Cardiac Muscle Arteriogenesis in Rats Immunofluorescence staining of sections from mouse skeletal A striking observation made in the analysis of histochemi- muscles transduced with the AAVs encoding PlGF or VEGF cally stained sections of TG rat, but not mouse, hearts was the revealed marked infiltration of CD45-positive leukocytes and presence of numerous large arteries of nearly the caliber of especially F4/80-positive macrophages. In contrast, overex- epicardial coronary arteries (Figure 3A and 3D, arrowheads). pression of either VEGF-B isoform could at best only mildly They were located deep in the myocardium and often close to increase macrophage infiltration (Figure IIIA and IIIB in the the endocardium, whereas similarly sized arteries were seen online-only Data Supplement). A similar difference was only on the epicardial side in WT hearts (Figure 3D, arrow). observed in parallel experiments in which the myocardium The total number of arteries as analyzed by Masson trichrome was transfected (Figure IIIC in the online-only Data staining for the adventitial layer was significantly increased Supplement). in TG heart sections at 4 months of age, especially in the subendocardial myocardium (Figure 3A through 3D). The Both Mice and Rats Overexpressing the Human number of arteries in TG rat hearts was significantly in- VEGF-B Gene in the Myocardium Exhibit creased also at 6 to 8 weeks of age (83Ϯ14 total arteries per Cardiac Hypertrophy heart section in ␣MHC–VEGF-B rat hearts versus 60Ϯ2 To test the therapeutic potential of VEGF-B both in mice and arteries per heart section in WT hearts; nϭ6 in both groups; in a larger animal model, we adopted a gain-of-function PϽ0.005). 1728 Circulation October 26, 2010

Table. Echocardiography of Cardiac Dimensions and Blood Pressure Analysis of Transgenic Rats and Wild-Type Controls

TG WT LVPWTd, mm 2.4Ϯ0.5‡ 1.8Ϯ0.5 IVSTd, mm 1.7Ϯ0.2 1.7Ϯ0.1 LV EF, % 83Ϯ8† 76Ϯ7 LV FS, % 48Ϯ12 41Ϯ9 ESV, mL 0.12Ϯ0.09† 0.19Ϯ0.07 SV, mL 0.55Ϯ0.14 0.59Ϯ0.08 EDV, mL 0.68Ϯ0.21 0.78Ϯ0.12 Heart rate, bpm 345Ϯ65 377Ϯ27 Heart rate, bpm* 237Ϯ38† 289Ϯ56 Diastolic BP, mm Hg 98Ϯ10 104Ϯ14 Systolic BP, mm Hg 138Ϯ12 144Ϯ16 LVPWTd indicates diastolic left ventricular posterior wall thickness; IVSTd, diastolic interventricular septal thickness; LV, left ventricle; EF, ejection fraction; FS, fractional shortening; ESV, end-systolic volume; SV, stroke volume; EDV, end-diastolic volume; and BP, blood pressure. *Under anesthesia. †PϽ0.05; ‡PϽ0.005; nϭ12 in each group.

vealed that PAI-1 was expressed mainly in the arterial walls (Figure 4C). Resorcin Fuchsin staining for elastin indicated that most of the smooth muscle cell proliferation had oc- curred in the media of the arteries of 1-month-old VEGF-B TG hearts, whereas some fragmentation of the internal elastic lamina was seen in many of the arteries from 4-month-old TG rats (Figure 3E, white arrowheads). In line with the results from the AAV-transfected mice, immunostaining with anti- bodies against the macrophage marker ED-1 did not indicate significant inflammatory cell infiltration (39Ϯ14 cells in ␣MHC–VEGF-B versus 36Ϯ9 cells in WT heart sections; PϾ0.05, nϭ5 in both groups). No increase in fibrosis could be seen in Masson trichrome staining of the hearts at 1 or 4 months (Figure 3D and data not shown). Figure 2. Cardiac hypertrophy in ␣MHC–VEGF-B mice and rats. A, Quantification of ratios of heart to body weight (HW/BW) in Immunofluorescent SMA staining confirmed the presence the ␣MHC–VEGF-B TG and WT rats (nϭ13 in both groups) and of a thick periendothelial smooth muscle cell layer in the TG mice (TG, nϭ10; WT, nϭ12). B, Representative dystrophin- arteries (Figure 4A and B). In contrast, staining for the rat stained sections of ␣MHC–VEGF-B rat and control rat hearts. Scale barϭ100 ␮m. C, Quantification of cardiomyocyte areas endothelial cell antigen RECA-1 did not reveal increased (nϭ5 in all groups). D, Echocardiographic analysis of the rat left blood capillary density (immunostained area, 11.7Ϯ2.2% in ventricular posterior wall thickness in diastole (LVPWTd; nϭ12 TG [nϭ4] versus 11.8Ϯ4.9% in WT hearts [nϭ5]; PϾ0.05). ␣ in each group). E, Blood pressure analysis of the MHC– 12 VEGF-B and control rats (nϭ12 in each group). F, Ratios of However, similar to previous observations in TG mice, we ␣ ϭ heart to body weight of MHC–VEGF-BEx1–5 (n 3) mice and WT could observe some increase in mean capillary vessel luminal littermates (nϭ4). *PϽ0.05; **PϽ0.005; ***PϽ0.0005. areas (data not shown). To analyze whether the myocardial arteries in the TG The average arterial vessel area was significantly increased hearts were functional and connected to the coronary arterial in the TG heart sections (Figure 3C), and the arterial tree, we performed contrast-enhanced micro–computed to- phenotype was also associated with dilation of the epicardial mography angiography of the rat hearts. The coronary arteries veins (Figure 3A, arrows). Consistent with the arteriogenesis, and veins and the epicardial vessels were found in anatomi- VEGF-B induced a number of cytokines that have been cally normal locations, but remarkably, the number and size associated with angiogenesis and tissue remodeling. Notably, of coronary vessel branches were strikingly increased in TG VEGF-B increased the expression of the delta-like 4 (Dll4) hearts compared with controls (Figure 5, arrows), with large notch ligand, which has been associated with arteriogenesis18 vessels also seen deep inside the myocardium (Figure 5, (Figure VI in the online-only Data Supplement). We also saw arrowhead, and data not shown). In vivo positron emission the induction of plasminogen activator inhibitor (PAI)-1, tomography showed homogeneous uptake of [11C]acetate which is involved in matrix remodeling and, for example, throughout the left ventricular myocardium of the rats. The myoendothelial junction formation.19 Immunostaining re- average myocardial perfusion was comparable in the TG and Bry et al VEGF-B as a Coronary Growth Factor 1729

Figure 3. The VEGF-B transgene induces strong arteriogenesis in rat myocardium. A, Representative Herovici-stained (HERO) sec- tions of 4-month-old rat hearts. Arrowheads indicate arteries; arrows indicate enlarged epicardial veins. B, Quantification of the total (top bars) and subendocardial (bottom bars) arterial vessel number in VEGF-B–overex- pressing rats (nϭ11) and controls (nϭ10) quantified with the aid of Masson trichrome (TRI) staining of the vessel adventitia. *PϽ0.05; **PϽ0.005. C, The average arterial vessel area in VEGF-B–overexpressing rats (nϭ11) and controls (nϭ10). D, Higher- magnification heart sections stained with Herovici and Masson trichrome. Arrowheads indicate arteries in TG rats compared with a similarly sized epicardial artery in a WT rat (arrow). Scale barϭ400 ␮m. Asterisks indi- cate the left ventricle. E, Representative images of Resorcin Fuchsin–stained elastin in heart sections at 1 and 4 months of age. White arrowheads indicate fragmentation of the internal elastic lamina of subendocardial arteries in the 4-month-old TG heart. The asterisk indicates the left ventricle. Scale barϭ200 ␮m.

Ϯ Ϯ ϭ Ϯ Ϫ Ϫ ϭ Ͼ WT rats (K1, 5.1 0.4 versus 5.0 0.7 per minute; n 4in 0.48 0.04 in VEGFR-1 TK / mice [n 4]; P 0.5), indi- each group; Pϭ0.86). TG and WT rats also demonstrated cating that VEGF-B does not induce hypertrophy in mice comparable myocardial oxygen consumption (Kmono, lacking the VEGFR-1 tyrosine kinase domain. However, as 0.50Ϯ0.11 versus 0.58Ϯ0.06 per minute; Pϭ0.24) and effi- previously observed, the ratios of heart to body weight of Ϫ Ϫ ciency of work (154Ϯ71 versus 138Ϯ42 mm Hg ⅐ L 1 ⅐ g 1; VEGF-B–overexpressing VEGFR-1–WT mice were in- Pϭ0.69). creased very significantly compared with non-TG VEGFR- 1–WT mice (heart to body weight, 0.53Ϯ0.04% in ␣MHC– ϭ Ϯ ϭ Neuropilin-1 Does Not Transduce VEGF-B Signals VEGF-B167 (n 10) versus 0.42 0.03% in WT (n 7) mice; Essential for the Hypertrophic Response PϽ0.0005). These results indicate that the VEGFR-1 path- To establish the VEGF-B signal transduction pathway in- way is essential for transduction of the hypertrophic signals volved in the hypertrophy response, we first concentrated on of VEGF-B. the 2 receptors of VEGF-B in mice. A C-terminally truncated form of VEGF-B containing the first 5 exons and thus lacking the isoform-specific sequences and therefore also the neuro- Loss of Bmx Reduces VEGF-B–Induced pilin-1– and heparin-binding domain15 was expressed in Cardiac Hypertrophy ␣ The cytoplasmic bone marrow kinase in X mouse heart ( MHC–VEGF-BEx1–5; Figure IVB and IV in the online-only Data Supplement). Analysis of the TG hearts (Bmx) is expressed in the arterial endothelium and myeloid indicated that they undergo a degree of hypertrophy similar to cell lineage; its deletion from mice does not lead to any that of hearts overexpressing the full-length gene or the apparent phenotype,21 but interestingly protects mice from Ͻ 22 VEGF-B167 form (P 0.05, Mann-Whitney U test; Figure cardiac hypertrophy induced by aortic constriction. To 2F). This indicated that, despite its expression at least in fetal evaluate whether Bmx deficiency inhibits the VEGF-B– myocardium,15 neuropilin-1 binding does not mediate the induced hypertrophy, we crossed Bmx gene–targeted mice ␣ hypertrophic response. with the MHC–VEGF-B167 mice. We then compared mice ␣ ␣ of 4 genetic backgrounds: MHC–VEGF-B167, MHC– Deletion of the VEGFR-1 Tyrosine Kinase Domain VEGF-B167;Bmx-/0, WT, and Bmx-/0 (KO) mice. At the age Attenuates the Hypertrophic Effect of VEGF-B of 3 to 3.5 months, when the hypertrophic phenotype of ␣ 12 To analyze the role of VEGFR-1 signal transduction in the MHC–VEGF-B167 mice is apparent, the mice were eutha- VEGF-B–induced cardiac hypertrophy, we mated the ␣MHC– nized and the heart tissues analyzed. Loss of Bmx attenuated the cardiac hypertrophy induced by VEGF-B (Figure 6A and VEGF-B167 TG mice with mice having a deletion of the VEGFR-1 tyrosine kinase domain (VEGFR-1 TKϪ/Ϫ) but no obvi- 6C). In addition, the average cardiomyocyte area was signif- ␣ ous phenotype.20 Analysis of offspring 3 to 4 months after icantly reduced in the MHC–VEGF-B167;Bmx KO hearts ␣ birth revealed that cardiac overexpression of VEGF-B had compared with the MHC–VEGF-B167 hearts (Figure 6B and no effect on ratios of heart to body weight in VEGFR-1 6D). These results indicate that Bmx mediates at least some TKϪ/Ϫ mice (heart to body weight, 0.49Ϯ0.04% in ␣MHC– of the signals for the VEGF-B–induced hypertrophy via the Ϫ Ϫ ϭ VEGF-B167;VEGFR-1 TK / [n 8] mice versus arterial endothelium. 1730 Circulation October 26, 2010

Figure 4. Immunofluorescence staining of thick cardiac sections. A, Tile scan images of thick cardiac sections stained for SMA. Arrow- heads indicate large subendocardial arteries in VEGF-B TG hearts. Scale barϭ1 mm. LV indicates left ventricle; RV, right ventricle. B, Representative images of SMA- and RECA-1–stained cardiac sections. Scale barϭ200 ␮m. C, Representative images of SMA- and PAI-1–stained cardiac sections. Arrows indicate colocalization of SMA and PAI-1. Scale barϭ200 ␮m.

Discussion found to elicit a profile typical of the In the present study, we show that VEGF-B induces a striking compensatory, hypertrophic response both in cultured cardio- growth of the coronary vascular tree in the rat, but not mouse, myocytes and in infarcted hearts.24 In mice, the hypertrophy heart and promotes myocardial hypertrophy in both species. ultimately resulted in exhaustion of triglycerides and accu- Importantly, VEGF-B, unlike PlGF or VEGF, did not signif- mulation of toxic lipid species, resulting in mitochondrial icantly increase vascular permeability or inflammatory cell autophagy/lysis and cardiomyopathy (Reference 12 and our recruitment into cardiac or skeletal muscle. The VEGF-B– unpublished observations). The cardiomyocytes in ␣MHC– induced hypertrophy did not compromise myocardial con- VEGF-B rats, however, have not shown signs of lipotoxic tractile function at least in rats up to 5 months of age, and the damage, perhaps because of a preserved perfusion of the impressive arteriogenesis was not associated with intimal hypertrophic myocardium induced by the strong coronary thickening, although the internal elastic lamina of the arteries arteriogenesis in this species. Importantly, functioning arter- had undergone some pathological changes in older rats. ies are essential for the sufficient perfusion of tissues.2 It has previously been shown that myocardial hypertrophy In contrast to VEGF-B, PlGF strongly increased vascular in the absence of other stimuli can be induced by angiogen- permeability. In addition, a novel snake venom VEGF has esis in mice.23 Heart-specific overexpression of both VEGF-B been shown to induce vascular permeability through prefer- isoforms in mice and rats reproduced the hypertrophic phe- ential signaling via VEGFR-1.25 Another major difference ␣ 12 notype previously seen in MHC–VEGF-B167 mice. Both observed between VEGF-B and PlGF was the lack of the TG mice and rats had significantly increased ratios of inflammatory cell recruitment into skeletal or cardiac muscle heart to body weight and hypertrophy of the cardiomyocytes. by VEGF-B. In previous work, the recruitment of bone Even though increases in capillary density were not observed marrow–derived cells, mainly monocytes/macrophages, cor- in the TG animals, it is conceivable that endothelium-derived related with arteriogenesis,2 and this required the neuropilin- functions activated by VEGF-B mediate the cardiac hyper- 1–binding domain of VEGF.26 trophy. Echocardiography confirmed the increased left ven- The hypertrophic effects of VEGF-B did not require tricular mass in the TG rats and, similar to our previous activation of neuropilin-1; overexpression of a truncated form analysis of mice, showed that despite circumferential hyper- of VEGF-B capable of binding to VEGFR-1 but lacking the trophy, VEGF-B overexpression did not compromise systolic neuropilin-1–binding domain15 induced significant cardiac function. Indeed, VEGFR-1 activation by VEGF-B has been hypertrophy. On the other hand, deletion of the VEGFR-1 Bry et al VEGF-B as a Coronary Growth Factor 1731

Figure 5. Representative 3-dimensional rendered micro–com- puted tomography images of vessels in VEGF-B TG rats and controls. Arrows indicate corresponding branching points in the TG and control hearts; arrowhead, large artery deep inside the myocardium of a TG rat. tyrosine kinase domain in the VEGF-B TG mice inhibited hypertrophy induced by VEGF-B, indicating that VEGFR-1 signaling provides the major pathway. This finding is of interest because the VEGFR-1 tyrosine kinase function seems dispensable for normal vascular development.27 Strikingly, in the TG rats but not in the mice, VEGF-B induced a remarkable growth of large arterial blood vessels that were found deep in the subendocardial myocardium, without causing capillary proliferation. These arteries were often larger than or equal in size to epicardial coronary arteries in control animals and were covered by a thick smooth muscle cell wall, with additional cell layers in the arterial media. Three-dimensional micro–computed tomogra- phy angiography suggested that the myocardial arteries Figure 6. Loss of Bmx tyrosine kinase reduces VEGF-B–in- formed part of an extended coronary arterial tree continuous duced cardiac hypertrophy. A, Representative ex vivo images of ␣ ϭ hearts from MHC–VEGF-B167 (VEGF-B TG/Bmx WT) (n 6), with the epicardial coronary vessels, which in the TG hearts VEGF-B TG/Bmx KO (nϭ21), WT (nϭ11), and Bmx KO (nϭ9) had larger and more numerous branches. The resting myo- mice. B, Representative images of dystrophin-stained cardiac cardial perfusion, oxygen consumption, and efficiency of sections. C, Quantification of ratios of heart to body weight. Val- Ϯ work were preserved in the TG rats. Thus, VEGF-B induced ues are presented as mean SEM. D, Quantification of cardio- myocyte areas. *PϽ0.05. Scale barϭ2 mm (A) and 20 ␮m (B). a strong arteriogenic response in the coronary vessels, which was also associated with the upregulation of Dll4 in the TG rat hearts. Importantly, this occurred without capillary angio- targeted mice indicated that their isolated hearts display genesis, vascular leakage, or inflammation. Increased PAI-1 vascular dysfunction after coronary occlusion and impaired 9 was detected in the TG rat hearts, which likely recovery from experimentally induced myocardial ischemia. resulted from the arterial growth because PAI-1 was ex- This finding and our present data warrant detailed analysis of pressed mainly in the arterial walls. A recent report has also the arteriogenic potential of VEGF-B in the rat heart in utero, indicated the importance of PAI-1 in myoendothelial junction during early postnatal coronary maturation, and in the adult formation.19 Further studies in animals with atherosclerosis or heart. It is also interesting to note that blockage of VEGFR-1 myocardial infarction are needed to demonstrate the potential signal transduction by use of a soluble form of the receptor beneficial effects of arteriogenesis on myocardial function. that captures VEGF, VEGF-B, and PlGF leads to myocardial We do not yet know why the human VEGF-B gene induces stunning or hibernation, which has been described in chronic 28 coronary arteriogenesis in rat but not mouse hearts. However, myocardial ischemia. On the other hand, VEGF-B167 was it is interesting that the results we obtained in the rats are reported to exert a powerful antiapoptotic effect on cardio- almost a mirror image of those reported by Bellomo et al9 in myocytes both in cell culture and in vivo after myocardial 24 the VEGF-B gene–targeted mice that developed slightly infarction. In addition, VEGF-B167 was also shown to smaller hearts than wild-type mice during the first postnatal preserve cardiac function in dogs developing tachypacing- month. In vitro perfusion experiments of the VEGF-B gene– induced dilated cardiomyopathy and to prevent oxidative 1732 Circulation October 26, 2010 stress and loss of mitochondrial membrane potential in Disclosures neonatal rat cardiomyocytes exposed to angiotensin II.29 Dr Alitalo is the principal investigator of research grants relevant These results suggest important cardioprotective roles for to the topic of this study from the Academy of Finland, Sigrid VEGF-B in heart failure. Juselius Foundation, and Helsinki University Hospital Funds. The other authors report no conflicts. Recent reports indicate that pressure overload–induced cardiac hypertrophy generated by thoracic aortic constriction References in mice is mediated at least in part through Bmx tyrosine 1. Gupta R, Tongers J, Losordo DW. Human studies of angiogenic gene kinase, which is expressed in the atrial endocardium and therapy. Circ Res. 2009;105:724–736. arterial endothelium.21,22 Furthermore, previous in vitro data 2. Schaper W. Collateral circulation: past and present. Basic Res Cardiol. 2009;104:5–21. from our laboratory show that Bmx phosphorylation can be 3. Ferrara N. Vascular endothelial growth factor: basic science and clinical stimulated by activated VEGFR-1, indicating a role in the progress. Endocr Rev. 2004;25:581–611. downstream signaling of this receptor.21 Here, we show that 4. Nagy JA, Benjamin L, Zeng H, Dvorak AM, Dvorak HF. Vascular perme- ability, vascular hyperpermeability and angiogenesis. Angiogenesis. 2008;11: Bmx deletion also reduces the cardiac hypertrophy induced 109–119. by VEGF-B overexpression. These results provide the first 5. Yla-Herttuala S, Alitalo K. Gene transfer as a tool to induce therapeutic evidence for the interaction between the VEGF-B and Bmx vascular growth. Nat Med. 2003;9:694–701. signaling pathways in vivo. Interestingly, the rescue of the 6. Lohela M, Bry M, Tammela T, Alitalo K. VEGFs and receptors involved in angiogenesis versus lymphangiogenesis. Curr Opin Cell Biol. 2009; phenotype in the Bmx KO background implicates the arterial 21:154–165. endothelium as an important source for the hypertrophic 7. Carmeliet P. VEGF gene therapy: stimulating angiogenesis or angioma- signals. However, further studies need to be performed to genesis? Nat Med. 2000;6:1102–1103. 8. Olofsson B, Pajusola K, Kaipainen A, von Euler G, Joukov V, Saksela O, improve our understanding of the mechanisms involved. Orpana A, Pettersson RF, Alitalo K, Eriksson U. Vascular endothelial The recent findings that VEGFR-1 activation by VEGF-B growth factor B, a novel growth factor for endothelial cells. Proc Natl increases cardiac mass and promotes maintenance of cardiac Acad Sci U S A. 1996;93:2576–2581. 9. Bellomo D, Headrick JP, Silins GU, Paterson CA, Thomas PS, Gartside contractility over time have obvious therapeutic implica- M, Mould A, Cahill MM, Tonks ID, Grimmond SM, Townson S, Wells tions.12,24,29 Prolonged cardiac hypertrophy leads to diastolic C, Little M, Cummings MC, Hayward NK, Kay GF. Mice lacking the insufficiency, whereas concurrent cardiac angiogenesis in vascular endothelial growth factor-B gene (Vegfb) have smaller hearts, such conditions has been shown to be important for preser- dysfunctional coronary vasculature, and impaired recovery from cardiac ischemia. Circ Res. 2000;86:E29–E35. 30,31 vation of cardiac function. VEGF-B may provide this 10. Aase K, von Euler G, Li X, Ponten A, Thoren P, Cao R, Cao Y, Olofsson essential function, as we demonstrate that overexpression of B, Gebre-Medhin S, Pekny M, Alitalo K, Betsholtz C, Eriksson U. both isoforms of VEGF-B in the rat heart induces concurrent Vascular endothelial growth factor-B-deficient mice display an atrial conduction defect. Circulation. 2001;104:358–364. myocardial growth and growth of arteries that penetrate into 11. Rissanen TT, Markkanen JE, Gruchala M, Heikura T, Puranen A, the myocardium. Importantly, VEGF-B does not promote Kettunen MI, Kholova I, Kauppinen RA, Achen MG, Stacker SA, Alitalo vascular leakage or tissue inflammation, both of which have K, Yla-Herttuala S. VEGF-D is the strongest angiogenic and lym- phangiogenic effector among VEGFs delivered into skeletal muscle via in part compromised previous attempts to use growth factors adenoviruses. Circ Res. 2003;92:1098–1106. of the VEGF family in therapeutic angiogenesis. Thus, 12. Karpanen T, Bry M, Ollila HM, Seppanen-Laakso T, Liimatta E, VEGF-B seems to be a more promising therapeutic candidate Leskinen H, Kivela R, Helkamaa T, Merentie M, Jeltsch M, Paavonen K, than PlGF or VEGF for patients with myocardial ischemia. Andersson LC, Mervaala E, Hassinen IE, Yla-Herttuala S, Oresic M, Alitalo K. Overexpression of vascular endothelial growth factor-B in mouse heart alters cardiac lipid metabolism and induces myocardial hypertrophy. Circ Res. 2008;103:1018–1026. Acknowledgments 13. Lahteenvuo JE, Lahteenvuo MT, Kivela A, Rosenlew C, Falkevall A, We thank Denis Tvorogov and Antti Nyka¨nen for professional Klar J, Heikura T, Rissanen TT, Vahakangas E, Korpisalo P, Enholm B, advice; Peter Andreasen for the PAI-1 antibody; and Tapio Tainola, Carmeliet P, Alitalo K, Eriksson U, Yla-Herttuala S. Vascular endothelial Katja Salo, Karita Viita-aho, Ulla Kiiski, Eeva Rouvinen, Riitta growth factor-B induces myocardium-specific angiogenesis and arterio- Sinervirta, Marita Heikkinen, Tuula Reponen, Marko Tirri, and Pa¨ivi genesis via vascular endothelial -1- and neuropilin Leinikka for technical assistance. We also thank the Molecular receptor-1-dependent mechanisms. Circulation. 2009;119:845–856. Imaging Unit at Biomedicum Helsinki and Fang Zhao for micro- 14. Olofsson B, Pajusola K, von Euler G, Chilov D, Alitalo K, Eriksson U. scope support and Nicolas Durant-Schaefers (GE Healthcare) for Genomic organization of the mouse and human for vascular endo- help with image processing. thelial growth factor B (VEGF-B) and characterization of a second splice isoform. J Biol Chem. 1996;271:19310–19317. 15. Makinen T, Olofsson B, Karpanen T, Hellman U, Soker S, Klagsbrun M, Sources of Funding Eriksson U, Alitalo K. Differential binding of vascular endothelial growth This study was supported financially by the Academy of Finland, factor B splice and proteolytic isoforms to neuropilin-1. J Biol Chem. Sigrid Juselius Foundation, Helsinki University Hospital Funds, 1999;274:21217–21222. Turku University Hospital Funds, Finnish Foundation for Cardio- 16. Tammela T, Enholm B, Alitalo K, Paavonen K. The biology of vascular vascular Research, and Centre of Excellence for Molecular Imaging endothelial growth factors. Cardiovasc Res. 2005;65:550–563. in Cardiovascular and Metabolic Research. M. Bry has personally 17. Fischer C, Mazzone M, Jonckx B, Carmeliet P. FLT1 and its ligands been supported by grants from Nylands Nation, Finska La¨karesa¨ll- VEGFB and PlGF: drug targets for anti-angiogenic therapy? Nat Rev Cancer. 2008;8:942–956. skapet, the Aarne Koskelo Foundation, Paulo Foundation, Emil 18. Liu ZJ, Shirakawa T, Li Y, Soma A, Oka M, Dotto GP, Fairman RM, Aaltonen Foundation, and Finnish Foundation for Cardiovascular Velazquez OC, Herlyn M. Regulation of Notch1 and Dll4 by vascular Research. Dr Kivela¨ has personally been supported by the Finnish endothelial growth factor in arterial endothelial cells: implications for Cultural Foundation and the Academy of Finland. Dr Holopainen has modulating arteriogenesis and angiogenesis. Mol Cell Biol. 2003;23: been supported by the Maud Kuistila Memorial Foundation, Finnish 14–25. Foundation for Cardiovascular Research, Aarne Koskelo Founda- 19. Heberlein KR, Straub AC, Best AK, Greyson MA, Looft-Wilson RC, tion, and Paulo Foundation. Sharma PR, Meher A, Leitinger N, Isakson BE. Plasminogen activator Bry et al VEGF-B as a Coronary Growth Factor 1733

inhibitor-1 regulates myoendothelial junction formation. Circ Res. 2010; induces vascular permeability through preferential signaling via VEGF 106:1092–1102. receptor-1. J Biol Chem. 2004;279:46304–46314. 20. Hiratsuka S, Minowa O, Kuno J, Noda T, Shibuya M. Flt-1 lacking the 26. Zacchigna S, Pattarini L, Zentilin L, Moimas S, Carrer A, Sinigaglia M, tyrosine kinase domain is sufficient for normal development and angio- Arsic N, Tafuro S, Sinagra G, Giacca M. Bone marrow cells recruited genesis in mice. Proc Natl Acad Sci U S A. 1998;95:9349–9354. through the neuropilin-1 receptor promote arterial formation at the sites of 21. Rajantie I, Ekman N, Iljin K, Arighi E, Gunji Y, Kaukonen J, Palotie A, adult neoangiogenesis in mice. J Clin Invest. 2008;118:2062–2075. Dewerchin M, Carmeliet P, Alitalo K. Bmx tyrosine kinase has a 27. Shibuya M. Vascular endothelial growth factor receptor-1 (VEGFR- redundant function downstream of angiopoietin and vascular endothelial 1/Flt-1): a dual regulator for angiogenesis. Angiogenesis. 2006;9: growth factor receptors in arterial endothelium. Mol Cell Biol. 2001;21: 225–230. 4647–4655. 28. May D, Gilon D, Djonov V, Itin A, Lazarus A, Gordon O, Rosenberger 22. Mitchell-Jordan SA, Holopainen T, Ren S, Wang S, Warburton S, Zhang C, Keshet E. Transgenic system for conditional induction and rescue of MJ, Alitalo K, Wang Y, Vondriska TM. Loss of Bmx nonreceptor chronic myocardial hibernation provides insights into genomic programs tyrosine kinase prevents pressure overload-induced cardiac hypertrophy. of hibernation. Proc Natl Acad Sci U S A. 2008;105:282–287. Circ Res. 2008;103:1359–1362. 29. Pepe M, Mamdani M, Zentilin L, Csiszar A, Qanud K, Zacchigna S, 23. Tirziu D, Chorianopoulos E, Moodie KL, Palac RT, Zhuang ZW, Tjwa Ungvari Z, Puligadda U, Moimas S, Xu X, Edwards JG, Hintze TH, M, Roncal C, Eriksson U, Fu Q, Elfenbein A, Hall AE, Carmeliet P, Giacca M, Recchia FA. Intramyocardial VEGF-B167 gene delivery Moons L, Simons M. Myocardial hypertrophy in the absence of external delays the progression towards congestive failure in dogs with pacing- stimuli is induced by angiogenesis in mice. J Clin Invest. 2007;117: induced dilated cardiomyopathy. Circ Res. 2010;106:1893–1903. 3188–3197. 30. Shiojima I, Sato K, Izumiya Y, Schiekofer S, Ito M, Liao R, Colucci WS, 24. Zentilin L, Puligadda U, Lionetti V, Zacchigna S, Collesi C, Pattarini L, Walsh K. Disruption of coordinated cardiac hypertrophy and angio- Ruozi G, Camporesi S, Sinagra G, Pepe M, Recchia FA, Giacca M. genesis contributes to the transition to heart failure. J Clin Invest. 2005; Cardiomyocyte VEGFR-1 activation by VEGF-B induces compensatory 115:2108–2118. hypertrophy and preserves cardiac function after myocardial infarction. 31. Sano M, Minamino T, Toko H, Miyauchi H, Orimo M, Qin Y, Akazawa H, FASEB J. 2010;24:1467–1478. Tateno K, Kayama Y, Harada M, Shimizu I, Asahara T, Hamada H, Tomita 25. Takahashi H, Hattori S, Iwamatsu A, Takizawa H, Shibuya M. A novel S, Molkentin JD, Zou Y, Komuro I. p53-induced inhibition of HIF-1 causes snake venom vascular endothelial growth factor (VEGF) predominantly cardiac dysfunction during pressure overload. Nature. 2007;446:444–448.

CLINICAL PERSPECTIVE Coronary heart disease is the leading cause of death in the Western world. Although pharmaceutical therapy and revascularization of stenotic coronary arteries are the standard therapy, many patients with advanced disease or small-vessel disease respond poorly to these treatments. Novel therapeutic strategies for promoting collateral artery formation, or arteriogenesis, are in high demand. Although vascular endothelial growth factor (VEGF) is the most potent angiogenic factor for possible therapy of myocardial ischemia, VEGF also promotes vascular leakage and inflammation, hampering its therapeutic utility. Our present findings show that VEGF-B, a hitherto poorly understood member of the VEGF family, acts as a coronary growth factor in the rat heart, stimulating the growth of epicardial vessels and large arteries deep inside the myocardium. Because cardiomyocyte damage in myocardial ischemia begins in the subendocardial myocardium, the VEGF-B–induced increased arterial supply to this area could have great therapeutic potential in ischemic heart disease. VEGF-B also does not cause increased vascular permeability or inflammatory cell recruitment, making it a promising candidate for therapeutic translation.